Multi-parameter photoelectric data fitting for microfluidic sweat colorimetric analysis

Abstract

Colorimetric detection has been widely used in sweat colorimetric analysis due to its advantages of convenient detection and easy implementation. However, its relatively low detection accuracy limits its application. It is therefore necessary to develop new methods to extract quantitative data from the colorimetric biosensor. In this work, a plug type paper-based analytical device (μPad) sensor was fabricated and put into a Polydimethylsiloxane (PDMS) microfluidic chamber prepared through 3D-Printing. Photoelectric sensor with a light-emitting diode (LED) light source was utilized for acquiring the colour change information (including R, G, B, Lux, colour temperature value) before and after the colorimetric sensing. Wireless Bluetooth transmission module was used for data acquisition. The obtained multi-parameter data was fitted through machine learning algorithm using Python to establish the relationship between the multi-parameter and the concentration of analyte. Discoloration experiments of glucose, lactic acid and pH were used to verify the feasibility of the sensor and data analysis method. Analysis of human sweat in a running volunteer has proved its potential as a wearable device and provided a new insight for data analysis of colorimetric method of wearable colorimetric biosensors.